Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Google fires another AI researcher who reportedly challenged findings

engadget.com

Post reply on HN
Print view View original

41–50 of 238 posts

Re: Google fires another AI researcher who reportedly challenged findings

#41

Post by mikkergp »

hn_throwaway_99 wrote: post #22

Obviously don't know the full context of why Satrajit Chatterjee was fired, but I can say that I followed the Timnit Gebru story quite closely, and there was a gigantic chasm, IMO, between what actually happened (and by "what actually happened" I mean that I'm conservatively just considering Gebru's own statements and tweets) and how things were reported in the wider press. The press will always be incentivized to re…

In the David vs Goliath battle David is incentivized to do exactly what Google is accusing this researcher of. Put out a controversial finding that supports your beliefs, regardless of whether or not it has been fully vetted. As a David you may not have the resources to fully vet the story anyways, particularly when the vetting process is defined by Goliath.

Goliath of course has as an ocean of toxic incentives to maintain the status quo, promote their own Agenda, and Gatekeep David. I understand why researchers want to work at these giant institutions. It's harder to influence the narrative if these giant institutions can operate unhindered. On top of that as you mention the press is incentivized to side with David since it's a more compelling story.

It 's part of the challenge of analyzing incentives, and means most people come to conclusions based on their prior beliefs. To come to the right conclusions you have to be able to dig deeper in the story than we have information to do so.

Re: Google fires another AI researcher who reportedly challenged findings

#42

Post by 21723 »

oh_sigh wrote: post #11

Were they fired because they challenged the findings, or were they fired for their behavior after being told their paper wasn't up to snuff for some reason? Same thing with Timnit...was she fired(or "resignation accepted") because of a paper she was writing, or was she fired because of her behavior after being told the paper needed more review(or whatever)? I really doubt that Google is so wrapped up in the "computer…

Scientists are pretty used to the idea of having their papers rejected. It happens all the time for all sorts of reasons. Never say never, but the likelihood that someone who's already gone through a PhD program would act out in some extreme way is low.

Usually, these terminations have more to do with "He's getting close to something untouchable" than anything else. It happens to journalists, when they do their job a bit too well, also.

Re: Google fires another AI researcher who reportedly challenged findings

#43

Post by 21723 »

The really brazen thing here is that Google is claiming he was fired "with cause".

People get let go all the time: layoffs, management changes, redundnacies. Usually, the company and manager try to avoid any legal or PR risk by saying nothing and letting the departing employee own the story. The fact that they're not doing this--that they're going out of their way to 9/11 someone's reputation after he's gone--suggests that someone in power is really, really pissed off.

Re: Google fires another AI researcher who reportedly challenged findings

#44

Post by 21723 »

ghaff wrote: post #30
oh_sigh wrote: post #11

Were they fired because they challenged the findings, or were they fired for their behavior after being told their paper wasn't up to snuff for some reason? Same thing with Timnit...was she fired(or "resignation accepted") because of a paper she was writing, or was she fired because of her behavior after being told the paper needed more review(or whatever)? I really doubt that Google is so wrapped up in the "computer…

Publicly "fired for cause" would seem like a pretty spectacular overreaction to someone having an academic disagreement with ML vs. human performance in some domain of chip design. Sufficiently spectacular that I'm essentially certain there's more to the story. Timnit was at least in a more... fraught area.

This. People get let go all the time and that's just how the game works, but if a company is publicly saying someone was "fired for cause", that means someone who outranks the lawyers is out to rape this guy's name, because that almost never happens otherwise.

Re: Google fires another AI researcher who reportedly challenged findings

#45

Post by 21723 »

hn_throwaway_99 wrote: post #22

Obviously don't know the full context of why Satrajit Chatterjee was fired, but I can say that I followed the Timnit Gebru story quite closely, and there was a gigantic chasm, IMO, between what actually happened (and by "what actually happened" I mean that I'm conservatively just considering Gebru's own statements and tweets) and how things were reported in the wider press. The press will always be incentivized to re…

It sounds like you're alleging that Timnit Gebru (whom I've never met, but who seems like a fine person from what I have read) is "toxic" and that working with her would be "a nightmare" to you. Am I correct in inferring this?

Re: Google fires another AI researcher who reportedly challenged findings

#46

Post by yodon »

guhidalg wrote: post #39
yodon wrote: post #14

Earlier quoted context omitted.

You may want to spend less time reading anti-intellectual media and more time talking to people who actually do academic scientific research. There are many challenges in academic science, but it does a remarkably good job of not falling prey to the strawman risks you believe in. Corporate research is of course subject to the economic interests of the corporation. Academic research is not held back from publishing in…

Does it? I picked the first professor from MIT's CSAIL group (Hal Abelson), found their latest published paper ( https://www.erudit.org/en/journals/irrodl/2022-v23-n1-irrodl... ) and found that they are funded by a Chinese governmental organization. Tell me how the research isn't being influenced by the needs and desires of the funder?

MIT CSAIL is not hard up for funding, nor is Hal Abelson. They can choose who they want to take money from. Funding sources that restrain a researcher's ability to publish damage that researcher's ability to advance their own career.

Many people here on HN have, for example, an irrational hatred and distrust of Microsoft. Microsoft Research, however is one of the biggest sources of computer science research funding in the world. If MSR funds you, they expect your work to result in publications and the primary factor in whether you get additional funding is whether your papers impact your field. Those characteristics make MSR funding highly desirable, precisely because it comes with no strings attached.

Re: Google fires another AI researcher who reportedly challenged findings

#48

Post by dekhn »

guhidalg wrote: post #39
yodon wrote: post #14

Earlier quoted context omitted.

You may want to spend less time reading anti-intellectual media and more time talking to people who actually do academic scientific research. There are many challenges in academic science, but it does a remarkably good job of not falling prey to the strawman risks you believe in. Corporate research is of course subject to the economic interests of the corporation. Academic research is not held back from publishing in…

Does it? I picked the first professor from MIT's CSAIL group (Hal Abelson), found their latest published paper ( https://www.erudit.org/en/journals/irrodl/2022-v23-n1-irrodl... ) and found that they are funded by a Chinese governmental organization. Tell me how the research isn't being influenced by the needs and desires of the funder?

Taiwan governmental organization (note, the coauthor who received the funding works in Taiwan). Details matter- Taiwan and China are two completely different entities.

Re: Google fires another AI researcher who reportedly challenged findings

#49

Post by nickdothutton »

Business is sometimes a difficult place to do pure research, just as a University is a very poor home for an operating business (outside of education). If you are a business who thinks there is some value in pure research then you could do worse than follow the model from the 50s and 60s. Separate research campus or unit. Separate management. Separate culture and working practices.
Post reply on HN